In this chapter he added some more modification. These are following:
- Giving eye (Ability to detect shape/
contour to be precise) He talked about detecting the shape that is in upright position with threshold activation and went on to expand on it. (A world populated with vehicles) - And this might not be that different from the principle in biology, how orchids, adopt bilaterally symmetrical shapes in order to be liked by insects.
- Depending on their programming, vehicles can respond differently to various shapes: i. avoid edge to be saved from fall ii. Line tracking
- Like in chapter 7 it will learn to associate certain shapes to certain outcome (isn’t it how Comma.ai try to solve the self driving car? with their
Openpilotsimulation)
This just goes on to show how modular approach can actually give us huge leap. I mean… This does seem like the vehicles have some 👀understanding 👀. But well… What is even understanding?
At least, I can say this is a visual processing. Right? And to me a simpleton this is understanding. May be if we animate such world. And add some more modularity for instance - communication, search for meaning… This world will look lot like ours.
11.01.25 -Abrar Matrikel-Nr: 5012300
(apow(e,1/(pow(x-b,2)/(2pow(c,2))))
(a*(1/pow(e,((pow(x-b,2)/(2*pow(c,2)))))))
x**2 => x^2
pow(x,2)
x**(-2) => x^(-2)
pow(x,-2) // Wroong
1/pow(x,2) // Correct